Improving the Knowledge Gradient Algorithm
Authors: Le Yang, Siyang Gao, Chin Pang Ho
NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 5 Numerical Experiments. In this section, we show empirical performances of the i KG, i KG-ϵ and i KG-F algorithms on synthetic and real-world examples. ... Tables 1-3 show the PFS of the algorithms under some fixed sample sizes (additional numerical results about the PFS and sampling rates of the tested algorithms are provided in the Supplement). The proposed i KG, i KG-ϵ and i KG-F perform the best. |
| Researcher Affiliation | Academia | Le Yang Department of Systems Engineering City University of Hong Kong lyang272-c@my.cityu.edu.hk Siyang Gao Department of Systems Engineering City University of Hong Kong siyangao@cityu.edu.hk Chin Pang Ho School of Data Science City University of Hong Kong clint.ho@cityu.edu.hk |
| Pseudocode | Yes | Algorithm 1: KG Algorithm; Algorithm 2: i KG Algorithm; Algorithm 3: i KG-ϵ Algorithm; Algorithm 4: i KG-F Algorithm |
| Open Source Code | No | No explicit statement or link providing concrete access to the source code for the described methodology was found. |
| Open Datasets | Yes | Dose-Finding Problem. We use the data in [28] (see ACR50 in week 16) for treating rheumatoid arthritis by the drug secukinumab. ... Drug Selection Problem. We consider five contraceptive alternatives based on the Drug Review Dataset (https://doi.org/10.24432/C5SK5S): ... Caption Selection Problem. We aim to select good captions based on the New Yorker Cartoon Caption Contest Dataset (https://nextml.github.io/caption-contest-data/). |
| Dataset Splits | No | The paper discusses a 'fixed-budget BAI' problem, where a total number of samples (budget) is used sequentially. It does not describe standard training, validation, and test dataset splits in the context of data partitioning for model evaluation. |
| Hardware Specification | No | No specific hardware details (like GPU/CPU models, memory, or detailed computer specifications) used for running experiments were provided. |
| Software Dependencies | No | No specific software dependencies with version numbers were mentioned in the paper. |
| Experiment Setup | Yes | For the tested algorithms, probabilities of false selection (PFS) are obtained based on the average of 100 macro-replications. ... We set the parameter β in TTEI as its default value 1/2. ... We set the input tolerance parameter as 0.0001 and the threshold as the posterior mean of the estimated best arm minus ϵ. ... We set the input tolerance parameter as 0.0001 and hyperparameter a = 25/H, where H is a constant that can be computed. |